AI RAG on K8s with NVIDIA & Modernizing a Policy Platform

Join us at Station Austin for Food, Drinks, and Kubernetes on Mar 26, 2026!

Everyone is invited to join! We aim to provide the latest news about Kubernetes and CNCF projects. 

We have three special guests: Geoff Niehaus (Executive Director of Software Engineering, CVS Health) will talk about AI RAG Agent on K8s with NVIDIA GPU; Joyjit Roy (Lead Principal Technical Program Manager, KForce) and Samaresh Kumar Singh (Principal Engineer, HP) will talk about Legacy to Cloud-Native: Modernizing a Policy Platform on K8s.

We will meet in Downtown Austin for the March episode.

Station Austin (formerly Capital Factory) is generously hosting the Kubernetes Austin meetup!

Address: 701 Brazos St, Austin, TX 78701

Upon arrival at the building, please proceed to the 1st floor.

We'll be in the Apollo Room (look for Capital Factory signs) after 5:45 pm.

We know parking in Downtown is tricky!

So, you can park in the building garage for just $8.00 (validation parking tickets will be distributed)! 

Street parking will still be an option. More information on parking here: https://www.capitalfactory.com/parking/

Venue Sponsored by Station Austin

Thank you to Station Austin for sponsoring Kubernetes Austin! Station Austin is the center of gravity for entrepreneurs in Texas. They meet the best entrepreneurs in Texas and introduce them to their first investors, employees, mentors, and customers. To sign up for a Station Austin membership, click here.

AI RAG Agent on K8s with NVIDIA GPU

What if you could run your own AI infrastructure for a fraction of what cloud APIs cost?

This session explores the journey of building a production-grade AI-powered chatbot from the ground up on your own hardware, on your own terms.

We'll walk through the key decisions and tradeoffs involved in standing up a modern AI platform: choosing and serving large language models, implementing retrieval-augmented generation for grounded responses, orchestrating intelligent agents that can reason and use tools, and wiring it all together with a responsive streaming interface.

Along the way, we'll tackle the infrastructure challenges that come with running AI at the edge — GPU scheduling, persistent storage, service mesh networking container orchestration, and CI/CD automation. You'll see how open-source CNCF projects can be composed into a cohesive platform that rivals managed cloud offerings.

We'll also break down the economics with real cost comparisons between self-hosted inference and commercial API pricing, demonstrating where the crossover points are and when it makes sense to bring AI workloads in-house.

Whether you're a platform engineer curious about AI infrastructure, a developer looking to move beyond API wrappers, or a leader evaluating the build-vs-buy decision for your organization. This talk will give you a practical roadmap and the confidence to start building.

Legacy to Cloud-Native: Modernizing a Policy Platform on K8s

This session shares lessons from modernizing a large, regulated enterprise platform from a legacy, tightly coupled architecture to a Kubernetes-based cloud-native environment. The journey involved decomposing monolithic services, introducing containerized workloads, and establishing a platform foundation that balanced developer autonomy with enterprise governance, security controls, and audit requirements.

We will cover practical design decisions around cluster architecture, workload isolation, CI/CD standardization, policy enforcement, and observability at scale. The talk also discusses what did not work, including early assumptions about service boundaries, release coordination across multiple teams, and friction with platform adoption.

From a business perspective, the modernization enabled faster release cycles, improved platform reliability, better resource utilization, and reduced operational risk in a compliance-heavy domain. Attendees will gain concrete patterns, architectural tradeoffs, and operational lessons they can apply when modernizing critical enterprise systems on Kubernetes.